Articles. Funding AstraZeneca.

Transcription

1 Ceftazidime-avibactam versus meropenem in nosocomial pneumonia, including ventilator-associated pneumonia (REPROVE): a randomised, double-blind, phase 3 non-inferiority trial Antoni Torres, Nanshan Zhong, Jan Pachl, Jean-François Timsit, Marin Kollef, Zhangjing Chen, Jie Song, Dianna Taylor, Peter J Laud, Gregory G Stone, Joseph W Chow Summary Background Nosocomial pneumonia is commonly associated with antimicrobial-resistant Gram-negative pathogens. We aimed to assess the efficacy and safety of ceftazidime-avibactam in patients with nosocomial pneumonia, including ventilator-associated pneumonia, compared with meropenem in a multinational, phase 3, double-blind, non-inferiority trial (REPROVE). Methods Adults with nosocomial pneumonia (including ventilator-associated pneumonia), enrolled at 136 centres in 23 countries, were randomly assigned (1:1) to 2000 mg ceftazidime and 500 mg avibactam (by 2 h intravenous infusion every 8 h) or 1000 mg meropenem (by 30-min intravenous infusion every 8 h) for 7 14 days; regimens were adjusted for renal function. Computer-generated randomisation codes were stratified by infection type and geographical region with a block size of four. Participants and investigators were masked to treatment assignment. The primary endpoint was clinical cure at the test-of-cure visit (21 25 days after randomisation). Non-inferiority was concluded if the lower limit of the two-sided 95% CI for the treatment difference was greater than 12 5% in the coprimary clinically modified intention-to-treat and clinically evaluable populations. This trial is registered with ClinicalTrials.gov (NCT ) and EudraCT ( ). Findings Between April 13, 2013, and Dec 11, 2015, 879 patients were randomly assigned. 808 patients were included in the safety population, 726 were included in the clinically modified intention-to-treat population, and 527 were included in the clinically evaluable population. Predominant Gram-negative baseline pathogens in the microbiologically modified intention-to-treat population (n=355) were Klebsiella pneumoniae (37%) and Pseudomonas aeruginosa (30%); 28% were ceftazidime-non-susceptible. In the clinically modified intention-to-treat population, 245 (68 8%) of 356 patients in the ceftazidime-avibactam group were clinically cured, compared with 270 (73 0%) of 370 patients in the meropenem group (difference 4 2% [95% CI 10 8 to 2 5]). In the clinically evaluable population, 199 (77 4%) of 257 participants were clinically cured in the ceftazidime-avibactam group, compared with 211 (78 1%) of 270 in the meropenem group (difference 0 7% [95% CI 7 9 to 6 4]). Adverse events occurred in 302 (75%) of 405 patients in the ceftazidime-avibactam group versus 299 (74%) of 403 in the meropenem group (safety population), and were mostly mild or moderate in intensity and unrelated to study treatment. Serious adverse events occurred in 75 (19%) patients in the ceftazidime-avibactam group and 54 (13%) patients in the meropenem group. Four serious adverse events (all in the ceftazidime-avibactam group) were judged to be treatment related. Interpretation Ceftazidime-avibactam was non-inferior to meropenem in the treatment of nosocomial pneumonia. These results support a role for ceftazidime-avibactam as a potential alternative to carbapenems in patients with nosocomial pneumonia (including ventilator-associated pneumonia) caused by Gram-negative pathogens. Funding AstraZeneca. Introduction Nosocomial pneumonia, which is also referred to as hospital-acquired pneumonia, is one of the most common hospital-acquired infections, 1 and is associated with high mortality and health-care expenditure. 2,3 Gram-negative pathogens particularly Pseudomonas aeruginosa and Enterobacteriaceae predominate in nosocomial pneumonia. 4 These pathogens often harbour several antimicrobial resistance mechanisms particularly extended-spectrum β-lactamases, and increasingly, carbapenemases. 5,6 Very few treatment options are available for infections caused by pathogens with extended-spectrum β-lactamases, and especially for those with carbapenemases. 7 Mortality risk and costs of treatment are increased in patients receiving inappropriate or delayed appropriate antibiotics. 8,9 Ceftazidime-avibactam combines the anti-pseudomonal cephalosporin ceftazidime, and the novel non-β-lactam Lancet Infect Dis 2017 Published Online December 15, S (17) See Online/Comment S (17)30748-X Servei de Pneumologia, Hospital Clinic, University of Barcelona, Barcelona, Spain (Prof A Torres MD); Institut D investigació August Pi I Sunyer, Barcelona, Spain (Prof A Torres); Ciber de Enfermedades Respiratorias, Spain (Prof A Torres); State Key Laboratory of Respiratory Disease, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China (Prof N Zhong MD); Charles University, Prague, Czech Republic (Prof J Pachl MD); APHP Hôpital Bichat-Claude Bernard, Paris-Diderot University, Paris, France (Prof J-F Timsit MD); Washington University School of Medicine, St Louis, MO, USA (Prof M Kollef MD); AstraZeneca, Shanghai, China (Z Chen MD, J Song MD); Taylormade Health, Warrington, United Kingdom (D Taylor BSc); Statistical Services Unit, University of Sheffield, Sheffield, UK (P J Laud MSc); Pfizer, Groton, CT, USA (G G Stone PhD); and Pfizer, Collegeville, PA, USA (J W Chow MD) Correspondence to: Prof Antoni Torres, Servei de Pneumologia, Hospital Clinic Barcelona, University of Barcelona, Barcelona, Spain atorres@clinic.cat Vol 376 August Published 28, online 2010 December 15,

2 Research in context Evidence before this study We searched PubMed with the terms randomized OR randomised AND ceftazidime AND avibactam OR ceftazidime AND avibactam AND Clinical Trial[ptyp] for articles published in English up to May 18, We identified two phase 2 studies and four phase 3 studies in which the efficacy and safety of ceftazidime-avibactam (with or without metronidazole as applicable for anaerobic coverage) was assessed in patients with serious Gram-negative infections, including complicated intra-abdominal infections (NXL104/2002 [NCT ], RECLAIM 1 and 2 [NCT ], RECLAIM 3 [NCT ]), complicated urinary tract infections (NXL104/2001 [NCT ], RECAPTURE 1 and 2 [NCT and NCT ]), or either complicated intra-abdominal or urinary tract infections caused by ceftazidime-non-susceptible Gram-negative pathogens (REPRISE [NCT ]). Across these studies, ceftazidime-avibactam had similar efficacy and safety to predominantly carbapenem comparators. These clinical data, and a phase 1 study (NCT ) showing that both ceftazidime and avibactam penetrate dose-proportionally into epithelial lining fluid, supported the clinical investigation of ceftazidime-avibactam in nosocomial pneumonia, including ventilator-associated pneumonia. Added value of this study The phase 3 REPROVE trial is the first clinical study, to our knowledge, of ceftazidime-avibactam in adults with nosocomial pneumonia (including ventilator-associated pneumonia). The patient population and pathogens isolated were consistent with those commonly observed in nosocomial pneumonia. Our results showed that ceftazidime-avibactam was non-inferior to meropenem, a standard therapy for nosocomial pneumonia, in this setting. Efficacy of ceftazidime-avibactam was similar against infections caused by ceftazidime-susceptible and ceftazidime-resistant pathogens. The safety profile of ceftazidime-avibactam was consistent with that previously noted with ceftazidime alone and with the known profile of ceftazidime-avibactam in patients with complicated intra-abdominal or urinary tract infections. Implications of all the available evidence Our results add to the evidence base showing the efficacy and safety of ceftazidime-avibactam in treating infections caused by Gram-negative pathogens, including those considered non-susceptible to ceftazidime, and support a role for ceftazidime-avibactam as a potential alternative to carbapenems in patients with nosocomial pneumonia (including ventilator-associated pneumonia) caused by Gram-negative pathogens. See Online for appendix β-lactamase inhibitor avibactam, which extends the in-vitro activity of ceftazidime to include Gram-negative organisms producing Ambler class A (eg, extended-spectrum β-lactamases, Klebsiella pneumoniae carbapenemase), class C (eg, AmpC), and some class D (eg, OXA- 48) β-lactamases This microbiological profile covers most carbapenem-non-susceptible Enterobacteriaceae and multidrug-resistant P aeruginosa (excluding metallo-βlactamase producers), and thus ceftazidime-avibactam is a potential alternative to carbapenems for the treatment of serious Gram-negative infections, including those caused by some carbapenemase-producing bacteria We did a randomised phase 3 trial to assess the non-inferiority of ceftazidime-avibactam to meropenem in terms of efficacy and safety in the treatment of nosocomial pneumonia. Methods Study design and participants REPROVE was a prospective, international, multicentre, parallel-group, randomised, double-blind, double-dummy, phase 3 non-inferiority trial. REPROVE was done at 136 centres (general hospitals) in 23 countries (appendix p22). The study was done in accordance with ethical principles that have their origin in the Declaration of Helsinki, and are consistent with the ICH Harmonised Tripartite Guideline E6(R1) for Good Clinical Practice, and applicable regulatory requirements. The study protocol and amendments (available at com) were approved by local ethics committees or institutional review boards. Eligible participants were aged years, were in hospital, and had acquired nosocomial pneumonia, which was defined as pneumonia with an onset 48 h or longer after admission or less than 7 days after discharge from an inpatient care facility. Ventilator-associated pneumonia was defined as parenchymal lung infection with an onset 48 h or longer after endotracheal intubation and mechanical ventilation. The term non-ventilatorassociated pneumonia was used to identify patients with nosocomial pneumonia who did not have ventilatorassociated pneumonia. Diagnosis of nosocomial pneumonia was based on clinical assessment, including new or worsening infiltrate on chest radiographs within 48 h of randomisation, and at least one systemic and two respiratory signs or symptoms of pneumonia. A respiratory specimen for Gram stain and culture was required within 48 h before randomisation. Key exclusion criteria included infections caused by any Gram-positive pathogens only or by other pathogens not expected to respond to ceftazidime-avibactam or meropenem, or both (polymicrobial infections were permitted if they included a target Gram-negative pathogen), and infections expected to require more than 14 days treatment. Patients without baseline culture data at randomisation could receive study therapy empirically. A full list of inclusion (pp 4 5) and exclusion criteria 2 Published online December 15, Vol 376 August 28, 2010

3 (pp 5 6) are in the appendix. All participants provided written informed consent. Randomisation and masking Investigators enrolled eligible patients, and then used an an interactive voice web response system to randomly assign them (1:1) to either ceftazidime-avibactam or meropenem. Patients were stratified by infection type (ie, ventilator-associated or non-ventilator-associated) and geographical region (western Europe, eastern Europe, China, rest of the world) at randomisation. Randomisation codes were computer-generated by AstraZeneca with the AstraZeneca Global Randomization System (block size of four). To maintain the blinding of study treatments, patients received double-dummy matching placebos (ie, ceftazidime-avibactam plus meropenem placebo or ceftazidime-avibactam placebo plus meropenem). Patients, investigators, and all study centre personnel were masked to study treatment, except for an unblinded pharmacist designee, who was responsible for maintaining accountability and preparing blinded study treatments. Procedures Patients assigned to ceftazidime-avibactam (AstraZeneca, Södertälje, Sweden) received a fixed-dose combination of 2000 mg ceftazidime and 500 mg avibactam by 2 h intravenous infusion every 8 h. Patients in the meropenem (ACS Dobfar, Milan, Italy) group received 1000 mg meropenem by 30 min intravenous infusion every 8 h. Dosages of both treatments were adjusted for patients with moderate or severe renal impairment (ie, creatinine clearance ml/min). After a protocol amendment (on Jan 9, 2015), ceftazidime-avibactam dosage adjustments in patients with renal impairment were modified (appendix pp 7 8), and the statistical analysis plan was updated to exclude patients with moderate or severe renal impairment at baseline who were randomly assigned before the protocol amendment from the main analyses to ensure the main efficacy and safety results were reflective of the approved ceftazidimeavibactam dosage regimens (data for excluded patients with moderate or severe renal impairment were summarised separately). Study treatment was discontinued after a minimum of 7 days (ie, 21 doses) and maximum of 14 days (ie, 42 doses). Patients awaiting identification of causative pathogens or susceptibility results from the baseline culture at randomisation received open-label linezolid or vancomycin for Gram-positive pathogen coverage. Open-label amikacin (or another aminoglycoside) for additional Gram-negative coverage was given to all patients awaiting baseline culture results for a minimum of h (extended depending on culture or susceptibility results) unless such treatment was contraindicated or patients were deemed at low risk of multidrug-resistant Gram-negative pathogens. Respiratory specimens for Gram stain and culture were obtained via endotracheal aspirate (ventilated patients), expectorated or induced sputum (non-ventilated patients), bronchoalveolar lavage, mini-bronchoalveolar lavage, or protected brush specimen at baseline, the endof- treatment visit (ie, within 24 h of the last dose of study treatment), and the test-of-cure visit (21 25 days after randomisation). Two sets of blood samples were collected from different sites for aerobic and anaerobic incubation within the 24 h before randomisation and as clinically indicated. If a previous culture was positive, repeat samples were taken at least every 3 days until bacteraemia cleared. Local laboratories did pathogen identification and susceptibility testing for all respiratory and blood isolates with Clinical Laboratory Standards Institute disk diffusion methods 16 against ceftazidime-avibactam, meropenem, and ceftazidime. Isolates identified by local laboratories and deemed pathogens by investigators were sent to a central reference laboratory for confirmation of identification and susceptibility testing. Patients had daily assessments from days 2 14, and at an end-of-treatment visit, a test-of-cure visit, and a final protocol follow-up visit days after randomisation (appendix pp 9 12). Clinical outcomes at the end-of-treatment and test-of-cure visits (appendix pp 13 14) were classified by investigators as cure (defined at the test-of- cure visit as resolution of all signs and symptoms of pneumonia such that no antibacterial therapy for nosocomial pneumonia was taken between the end-of-treatment and test-of-cure visits, inclusive), indeterminate, or treatment failure. Per-pathogen and per-patient microbiological responses (appendix p 15) were assessed as favourable (eradication or presumed eradication), unfavourable (persistence, persistence with increasing minimum inhibitory concentration, or presumed persistence), or indeter minate (per-pathogen responses only). Outcomes The primary endpoint was the proportion of patients clinically cured at the test-of-cure visit in the coprimary clinically modified intention-to-treat and clinically evaluable populations (appendix pp 16 17). The clinically modified intention-to-treat population comprised patients who met minimum disease criteria (on the basis of inclusion criteria; appendix pp 4 5) with one or more eligible Gram-negative pathogen, or those without any identifiable pathogen (patients with only non-target pathogens were excluded). The clinically evaluable population comprised patients in the clinically modified intention-to-treat population who received an adequate course of treatment and had an assessable clinical outcome within the assessment window, no protocol deviations that could affect the assessment of efficacy, and no unacceptable previous or concomitant antibiotics (appendix p 5). Secondary endpoints included clinical response at the end-of-treatment visit in the clinically modified intentionto-treat and clinically evaluable populations; clinical response at the end-of-treatment and test-of-cure visits in Vol 376 August Published 28, online 2010 December 15,

4 the microbiologically modified intention-to-treat, extended microbiologically evaluable, and microbiologically evaluable populations; all cause-mortality at the test-of-cure visit and at day 28 in the clinically evaluable and clinically modified and microbiologically modified intention-to-treat populations; clinical response at the end-of-treatment and test-of-cure visit in patients with ceftazidime-non-susceptible pathogens in the clinically evaluable, clinically modified intention-to-treat and microbiologically evaluable populations; and per-patient and per-pathogen microbiological responses at the end-of-treatment and test-of-cure visits in the microbiologically modified intention-to-treat, microbiologically evaluable, and extended microbiologically evaluable populations. A full list of secondary and exploratory analyses is in the appendix (pp 18 20). Safety assessments included monitoring of adverse events, clinical laboratory assessments, electro cardiograms, and mortality. Adverse events were summarised for events occurring from the first dose of study treatment to the final protocol follow-up visit. Adverse events occurring from the time when informed consent was obtained to the first dose of study treatment were recorded, but are not reported here. Statistical analyses The study was sized to ensure that power was sufficient (at least 85%) for the coprimary hypothesis tests against a 12 5% non-inferiority margin, in line with guidance from the European Medicines Agency. 17 We expected the underlying clinical cure rate at the test-of-cure visit to be around 78% in the clinically evaluable population and 70% in the clinically modified intention-to-treat population, which 50% and 85% of patients, respectively, would be eligible for inclusion in. The number of patients to be randomly assigned for the primary analysis for non-inferiority was approximately 790, to obtain around 394 and 670 evaluable patients in the clinically evaluable and clinically modified intention-to-treat populations, respectively. The estimated power with these numbers of assessable patients (with the previously described assumptions of cure rates and a one-sided α of 2 5%) was calculated with nquery (version 7) via the Newcombe- Wilson score method (uncorrected). 18 Patients with ventilator-associated pneumonia were recruited in parallel with other patients. Statistical analyses and the non-inferiority assessment for the primary endpoint were based on the difference in clinical cure rates between treatment groups. The safety population comprised all patients who received any amount of study therapy. An independent data monitoring committee was established with a charter to ensure that the safety of patients was not compromised. Two-sided 95% CIs for difference between treatments in the proportion of patients with clinical cure were computed with the unstratified method of Miettinen and Nurminen. 19 For primary efficacy endpoints, non-inferiority of ceftazidime-avibactam to meropenem was deemed to be shown if the lower limit of the two-sided 95% CI for the treatment difference (ceftazidime-avibactam minus meropenem) was greater than 12 5%, and the p value was calculated for the corresponding one-sided non-inferiority hypothesis test. Three sensitivity analyses were done for the primary efficacy variable: adjustment for the effect of prespecified stratification factors, type of infection (ie ventilatorassociated or non-ventilator-associated), and geographical region; analysis of patients who had received potentially effective concomitant antibiotics as having indeterminate clinical response at the test-of-cure visit; and analysis of patients who died after the test-of-cure visit and up to the final protocol follow-up visit as clinical failures at the testof-cure visit. Prespecified subgroup analyses assessed the effect of baseline patient and disease characteristics, including infection type, study region, Acute Physiology and Chronic Health Evaluation II (APACHE II) category, previous systemic antibiotic use, presence of bacteraemia, and baseline renal function (including moderate or severe impairment and augmented renal clearance [creatinine clearance >151 ml/min]). Concomitant aminoglycoside use was assessed as an exploratory posthoc subgroup analysis, with patients assigned to categories of concomitant aminoglycoside exposure defined before study database lock by blinded review of post-baseline data. All statistical analyses were done in SAS (version 9.1 or higher). This trial is registered with ClinicalTrials.gov (NCT ) and EudraCT ( ). Role of the funding source The study sponsor was involved in study design; data collection, analysis, and interpretation; and data checking of information provided in the Article. Responsibility for opinions, conclusions, and data interpretation lies with the authors. All authors had full access to all study data and final responsibility for the decision to submit for publication. Results Between April 13, 2013, and Dec 15, 2015, 879 patients were randomly assigned (table 1). After 62 patients with moderate or severe renal impairment (who were randomly assigned before the protocol amendment) were excluded from the main analyses, 409 patients were assigned to ceftazidime-avibactam and 408 to meropenem, of whom 405 and 403, respectively, received study treatment and comprised the safety population (figure 1). Baseline and disease characteristics were generally well balanced (table 1; appendix pp 76 77). Main reason for exclusion from the clinically modified intention-to-treat population was isolation of only Gram-positive pathogens at baseline (46 [11%] in the ceftazidime-avibactam group and 31 [8%] patients in the meropenem group). 70 (17%) in the ceftazidime-avibactam 4 Published online December 15, Vol 376 August 28, 2010

5 group and 61 (15%) in the meropenem group had important protocol deviations leading to exclusion from the clinically evaluable population; the main reasons for exclusion were receipt of concomitant non-protocolled antibiotics with potential effects on efficacy up to the testof-cure visit (43 (11%) in the ceftazidime-avibactam group and 46 [11%] in the meropenem group) and not having a response of cure or failure at the test-of-cure visit (40 [10%] in the ceftazidime-avibactam group and 37 [9%] in the meropenem group). Of 817 randomly assigned patients, 355 (44%) were included in the microbiologically modified intention-to-treat population. Baseline pathogens were similar between groups and as expected for patients with nosocomial pneumonia (appendix pp 23 26). Prominent Gram-negative pathogens isolated from respiratory site or blood were K pneumoniae and P aeruginosa (appendix p 23). 100 patients (28%) had one or more ceftazidime-non-susceptible Gram-negative pathogen, including 79 with Enterobacteriaceae and 25 with P aeruginosa. Staphylococcus aureus (detected in 58 patients [16%]) was the only Gram-positive pathogen to be isolated in ten or more patients. The minimum inhibitory concentrations required to inhibit the growth of at least 90% of isolates (MIC 90 ) with ceftazidime and ceftazidime-avibactam were greater than 32 mg/l and 0 5 mg/l, respectively, against 317 isolates of Enterobacteriaceae, and greater than 32 mg/l and 8 mg/l, respectively, against 101 isolates of P aeruginosa tested at the central laboratory. 79 (25%) Enterobacteriaceae isolates and 25 (24 8%) P aeruginosa isolates were non-susceptible to ceftazidime by Clinical Laboratory Standards Institute criteria (minimum inhibitory concentration >4 mg/l for Enterobacteriaceae and >8 mg/l for P aeruginosa). Thus, the ceftazidime-avibactam minimum inhibitory concentration distribution was left-shifted compared with that of ceftazidime alone (appendix p 84). Meropenem MIC 90 values against the same isolates were 0 12 mg/l for Enterobacteriaceae and greater than 8 mg/l for P aeruginosa. Two isolates of K pneumoniae and nine isolates of P aeruginosa were resistant to ceftazidime-avibactam, and six isolates of Enterobacteriaceae (five K pneumoniae and one Serratia marcescens) and 31 isolates of P aeruginosa were not susceptible to meropenem (minimum inhibitory concentration >2 mg/l). Overall, two K pneumoniae isolates and eight P aeruginosa were nonsusceptible to either study drug. Of the 355 patients in the microbiologically modified intention-to-treat population, 203 (57%) had monomicrobial infections and 152 (43%) had polymicrobial infections. 66 (19%) had a mixture of Gram-negative and Gram-positive pathogens. These results were balanced between treatment groups and were generally similar for the extended microbiologically evaluable and the microbiologically evaluable populations (data not shown). Ceftazidime-avibactam was non-inferior to meropenem in both coprimary analysis populations (figure 2). In the clinically modified intention-to-treat population, 245 (68 8%) of 356 people in the ceftazidime-avibactam Ceftazidime-avibactam (n=356) group were clinically cured at the test-of-cure visit, compared with 270 (73 0%) of 370 in the meropenem group (difference 4 2 [95% CI to 2 46]; p=0 0066). 199 (77 4%) of 257 in the ceftazidime-avibactam Meropenem (n=370) Age (years) 62 1 (16 6) 61 9 (17 4) Male sex 268 (75%) 274 (74%) Body-mass index (kg/m2)* (6 11) (5 17) Race White 150 (42%) 163 (44%) Black or African American 1 (<1%) 2 (1%) Asian 201 (56%) 199 (54%) Other 4 (1%) 6 (2%) APACHE II Score 14 5 (4 01) 14 9 (4 05) <10 1 (<1%) 1 (<1%) (87%) 316 (85%) (13%) 53 (14%) Renal status Estimated creatinine clearance (ml/min) (67 5) (53 1) Normal renal function or mild impairment 286 (80%) 292 (79%) ( ml/min) Moderate or severe impairment (16 50 ml/min) 18 (5%) 18 (5%) Augmented (>151 ml/min) 50 (14%) 58 (16%) Type of nosocomial pneumonia Ventilator-associated pneumonia 118 (33%) 128 (35%) Non-ventilator-associated pneumonia 238 (67%) 242 (65%) Type of ventilator-associated pneumonia infection Early 29 (8%) 47 (13%) Late 89 (25%) 81 (22%) Mechanical ventilation at baseline Ventilated 154 (43%) 159 (43%) Not ventilated 202 (57%) 211 (57%) Bacteraemic 19 (5%) 15 (4%) Infection type Monomicrobial 104 (29%) 105 (28%) Polymicrobial 69 (19%) 83 (22%) No study-qualifying pathogen identified 183 (51%) 182 (49%) Previous systemic antibiotic use None 122 (34%) 117 (32%) 24 h 185 (52%) 209 (56%) >24 to 48 h 49 (14%) 44 (12%) Concomitant aminoglycoside use None 72 (20%) 68 (18%) >0 to 72 h 199 (56%) 225 (61%) >72 h 85 (24%) 77 (21%) Data are mean (SD) or n (%). APACHE=Acute Physiology and Chronic Health Evaluation. *Data were missing for ten patients in the ceftazidime-avibactam group and nine patients in the meropenem group. As reported by the site from the Cockcroft-Gault method on the basis of local laboratory data; data were missing for two patients in each group. In the 48 h before randomisation. Exploratory analysis (not defined a priori in the clinical study protocol); the concomitant aminoglycoside subgroups are not based on a baseline patient characteristic, but were defined by blinded review of post-baseline data. Table 1: Baseline patient demographics and disease characteristics in clinically modified intention-to-treat population Vol 376 August Published 28, online 2010 December 15,

6 409 randomly assigned to ceftazidime-avibactam 405 received ceftazidime-avibactam 355 completed study and attended final protocol follow-up visit 879 patients randomly assigned 62 patients with moderate or severe renal impairment at baseline excluded 4 did not receive study treatment 1 patient decision 2 died 1 other reasons 405 in safety population 401 in MITT population 356 in cmitt population 257 in CE population 171 in mmitt population 125 in eme population 50 patients discontinued the study 8 patient decision 37 died 3 lost to follow-up 2 other reasons 408 randomly assigned to meropenem 403 received meropenem 363 completed study and attended final protocol follow-up visit 5 did not receive study treatment 1 did not meet eligibility criteria 1 died 3 other reasons 403 in safety population 401 in MITT population 370 in cmitt population 270 in CE population 184 in mmitt population 131 in eme population 39 patients discontinued the study* 4 patient decision 1 did not meet eligibility criteria 27 died 7 lost to follow-up 1 other reason Figure 1: Trial profile CE=clinically evaluable. cmitt=clinically modified intention-to-treat population. eme=extended microbiologically evaluable. MITT=modified intention-to-treat. mmitt=microbiologically modified intention-to-treat. *One patient in the meropenem group completed the test-of-cure visit (which was out of window) and the final protocol follow-up visit on the same day, and was treated as having neither completed nor discontinued the study. Clinically modified intention-to-treat population Clinically evaluable population Ceftazidime-avibactam vs meropenem 245 (68 8%) vs 270 (73 0%) 199 (77 4%) vs 211 (78 1%) Difference in clinical cure rate (95% CI) Favours meropenem Favours ceftazidime-avibactam Figure 2: Clinical cure rates at test-of-cure visit Data are number of patients with clinical cure (%). Dashed line indicates non-inferiority margin of 12 5%. group and 211 (78 1%) of 270 in the meropenem group were cured in the clinically evaluable population (difference 0 7 [95% CI 7 86 to 6 39]; p=0 0007). Similar results were noted in secondary analysis populations (appendix p 85). Results of sensitivity analyses that adjusted for stratification factors, or in which patients who died after the test-of-cure visit were deemed clinical failures at the test-of-cure visit, were consistent with those of the primary analysis (data not shown). In the clinically modified intention-to-treat population, 24 (7%) of 356 participants in the ceftazidime-avibactam group and 33 (9%) of 370 in the meropenem group with clinical cure at the test-of-cure visit received potentially effective concomitant antibiotics. Sensitivity analysis adjusted for treatment with potentially effective concomitant antibiotics (in the clinically modified intention-to-treat population) accounted for a 2 2% shift in treatment difference, with clinical cure rates of 221 (62 1%) in the ceftazidime-avibactam group and 237 (64 1%) in the meropenem group (difference 2 0 [95% CI 8 99 to 5 04]). Subgroup analyses of the primary endpoint (figure 3) showed no trends associated with various patient factors, including baseline renal status (including moderate or severe renal impairment and augmented renal function), previous systemic antibiotic use, type of infection (ie, non-ventilator-associated vs ventilator-associated and early vs late ventilator-associated pneumonia), and APACHE II score category. Cure rates were generally similar across treatment groups and in both coprimary populations in each subgroup. Clinical cure rates were similar across treatment groups in the exploratory analysis of patients who received concomitant aminoglycosides (either 72 h or >72 h; appendix pp 27 28) and those who did not. Per-pathogen clinical cure rates at the test-of-cure visit were generally similar between treatment groups, with numerical differences with wide CIs among individual bacterial species (table 2). Results of other secondary efficacy analyses are presented in the appendix (pp 29 72, 85). Per-pathogen clinical cure rates at the test-of-cure visit among patients infected with ceftazidime-nonsusceptible pathogens in the clinically evaluable population were similar between groups (29 [80 6%] of 36 in the ceftazidime-avibactam group vs 32 [78 0%] of 41 in the meropenem group; difference 2 5% [95% CI to 20 74]), and were also similar to those in patients in whom only ceftazidime-susceptible pathogens were isolated at baseline (63 [75 0%] of 84 vs 69 [78 4%] of 88; difference 3 4% [95% CI to 9 30]). Of the 62 patients with moderate or severe renal impairment at baseline, 58 were included in the clinically modified intention-to-treat population and 44 were included in the clinically evaluable population. At the test-of-cure visit, 18 (60%) of 30 patients in the ceftazidime-avibactam group and 16 (57%) of 28 in the 6 Published online December 15, Vol 376 August 28, 2010

7 Clinically modified intention-to-treat population Ceftazidime-avibactam vs meropenem Clinically evaluable population Ceftazidime-avibactam vs meropenem Region Western Europe 21 (72 4%) vs 19 (70 4%) 11 (73 3%) vs 11 (73 3%) Eastern Europe 76 (79 2%) vs 73 (76 8%) 62 (87 3%) vs 54 (84 4%) China Rest of the world 73 (58 9%) vs 91 (70 0%) 75 (70 1%) vs 87 (73 7%) 68 (66 0%) vs 80 (74 8%) 58 (85 3%) vs 66 (78 6%) APACHE II category (70 2%) vs 236 (74 7%) 182 (78 4%) vs 191 (79 9%) (58 7%) vs 34 (64 2%) 17 (68 0%) vs 20 (64 5%) Type of infection VAP 83 (70 3%) vs 95 (74 2%) 62 (77 5%) vs 63 (75 9%) Early VAP 20 (69 0%) vs 33 (70 2%) 14 (70 0%) vs 20 (71 4%) Late VAP 63 (70 8%) vs 62 (76 5%) 48 (80 0%) vs 43 (78 2%) Non-VAP 162 (68 1%) vs 175 (72 3%) 137 (77 4%) vs 148 (79 1%) Previous systemic antibiotics use in the 48 h before randomisation Yes 151 (64 5%) vs 176 (69 6%) 120 (75 5%) vs 133 (75 1%) 24 h >24 h No 124 (67 0%) vs 148 (70 8%) 27 (55 1%) vs 28 (63 6%) 94 (77 0%) vs 94 (80 3%) 107 (74 8%) vs 120 (75 9%) 13 (76 5%) vs 13 (68 4%) 79 (81 4%) vs 78 (83 9%) Failure of previous antibiotics use Yes No 15 (68 2%) vs 14 (63 6%) 218 (71 0%) vs 242 (74 2%) 13 (76 5%) vs 13 (68 4%) 186 (77 5%) vs 198 (78 9%) Type of microbial infection Monomicrobial 71 (68 3%) vs 78 (74 3%) 62 (75 6%) vs 65 (77 4%) Polymicrobial No study-qualifying pathogen identified 51 (73 9%) vs 64 (77 1%) 123 (67 2%) vs 128 (70 3%) 35 (79 5%) vs 41 (82 0%) 102 (77 9%) vs 105 (77 2%) Bacteraemic at baseline Yes 13 (68 4%) vs 9 (60 0%) 12 (70 6%) vs 7 (70 0%) No 232 (68 8%) vs 261 (73 5%) 187 (77 9%) vs 204 (78 5%) Ventilation status at baseline Ventilated 105 (68 2%) vs 114 (71 7%) 78 (75 7%) vs 76 (73 8%) Non-Ventilated 140 (69 3%) vs 156 (73 9%) 121 (78 6%) vs 135 (80 8%) Renal status Moderate or severe impairment (creatinine clearance ml/min) Normal or mild impairment (creatinine clearance ml/min) 13 (72 2%) vs 12 (66 7%) 187 (65 4%) vs 208 (71 2%) 11 (84 6%) vs 12 (70 6%) 156 (74 6%) vs 165 (77 1%) Augmented (creatinine clearance 151 ml/min) 43 (86 0%) vs 48 (82 8%) 31 (91 2%) vs 33 (86 8%) Aminoglycoside use* None >0 to 72 h >72 h 54 (75 0%) vs 49 (72 1%) 132 (63 3%) vs 164 (72 9%) 59 (69 4%) vs 57 (74 0%) 38 (80 9%) vs 29 (76 3%) 108 (77 1%) vs 134 (78 4%) 53 (75 7%) vs 48 (78 7%) Difference in clinical cure rate (95% CI) Favours Favours meropenem ceftazidime-avibactam Difference in clinical cure rate (95% CI) Favours Favours meropenem ceftazidime-avibactam Figure 3: Subgroup analysis of clinical cure rates at test-of-cure visit Data are number of patients with clinical cure (%) or number of patients in subgroup (%). APACHE II=Acute Physiology and Chronic Health Evaluation II. VAP=ventilator-associated pneumonia. *Exploratory analysis (not defined a priori in the clinical study protocol); all other subgroup analyses were prespecified in the study protocol. The concomitant aminoglycoside subgroups are not based on a baseline patient characteristic, and were defined before the study database lock and assigned by blinded review of post-baseline data. Vol 376 August Published 28, online 2010 December 15,

8 Patients with clinical cure (clinically evaluable population) Ceftazidimeavibactam (n=257) Meropenem (n=270) % difference (95% CI) Ceftazidimeavibactam (n=125) Patients with favourable microbiological response* (extended microbiologically evaluable population) Meropenem (n=131) % difference (95% CI) Enterobacteriaceae Klebsiella pneumoniae 31/37 (83 8%) 39/49 (79 6%) 4 2 ( to 20 50) 29/37 (78 4%) 39/49 (79 6%) 1 2 ( to 15 96) Enterobacter cloacae 20/21 (95 2%) 7/11 (63 6%) 31 6 (4 79 to 61 30) 18/21 (85 7%) 7/11 (63 6%) 22 1 ( 8 07 to 53 69) Escherichia coli 8/11 (72 7%) 14/18 (77 8%) 5 1 ( to 25 79) 10/11 (90 9%) 16/18 (88 9%) 2 0 ( to 26 44) Proteus mirabilis 11/11 (100 0%) 7/8 (87 5%) 12 5 ( to 48 07) 9/11 (81 8%) 6/8 (75 0%) 6 8 ( to 46 51) Serratia marcescens 10/12 (83 3%) 8/8 (100 0%) 16 7 ( to 19 48) 9/12 (75 0%) 5/8 (62 5%) 12 5 ( to 51 82) Enterobacter aerogenes 4/6 (66 7%) 2/5 (40 0%) 26 7 ( to 70 73) 5/6 (83 3%) 3/5 (60 0%) 23 3 ( to 68 33) Gram-negative pathogens other than Enterobacteriaceae Pseudomonas aeruginosa 27/42 (64 3%) 27/35 (77 1%) 12 8 ( to 8 01) 18/42 (42 9%) 14/35 (40 0%) 2 9 ( to 24 32) Haemophilus influenzae 10/11 (90 9%) 11/13 (84 6%) 6 3 ( to 36 09) 11/11 (100 0%) 12/13 (92 3%) 7 7 ( to 34 00) Gram-positive aerobes Staphylococcus aureus 11/14 (78 6%) 16/22 (72 7%) 5 8 ( to 32 67) 5/14 (35 7%) 17/22 (77 3%) 41 6 ( to 8 36) *Eradication or presumed eradication of the baseline pathogens. Table 2: Per-pathogen clinical cure rates and favourable microbiological response rates at test-of-cure visit meropenem group were clinically cured in the clinically modified intention-to-treat population, whereas in the clinically evaluable population, 15 (71%) of 21 patients in the ceftazidime-avibactam group and 13 (57%) of 23 in the meropenem group were clinically cured. All-cause mortality was similar across treatment groups at both the test-of-cure visit and day 28. In the clinically modified intention-to-treat population, 29 (8 1%) of 356 died in the ceftazidime-avibactam and 25 (6 8%) of 370 in the meropenem group died by the test-of-cure visit (difference 1 4 [95% CI 2 48 to 5 35]), whereas 30 (8 4%) and 27 (7 3%), respectively, died by day 28 (difference 1 1 [95% CI 2 84 to 5 18]). In the clinically evaluable population, 11 (4 3%) of 257 died in the ceftazidime-avibactam group and eight (3 0%) of 270 died in the meropenem group by the test-of-cure visit (difference 1 3 [95% CI 2 01 to 4 89]), whereas 12 (4 7%) and nine (3 3%), respectively, died by day 28 (difference 1 3 [95% CI 2 14 to 5 04]). Per-patient favourable microbiological response rates at the test-of-cure visit were generally lower than clinical cure rates, but were similar between the ceftazidimeavibactam and meropenem groups and consistent across the microbiologically modified intention-to-treat (95 [55 6%] of 171 vs 118 [64 1%] of 184; difference 8 6 [95% CI to 1 64]), extended microbiologically evaluable (80 [64 0%] of 125 vs 89 [67 9%] of 131; difference 3 9 [95% CI to 7 66]), and microbiologically evaluable (70 [65 4%] of 107 vs 83 [70 3%] of 118; difference 4 9 [95% CI to 7 28]) populations. In patients infected with ceftazidime-non-susceptible pathogens, perpatient favourable microbiological response rates were similar between groups at the end-of-treatment and testof-cure visits in the microbiologically modified intentionto-treat, extended microbiologically evaluable, and micro biologically evaluable populations (appendix p 67), and were similar to the overall per-patient favourable microbiological response rates. Favourable per-pathogen microbiological response (eradication or presumed eradication) rates at the test-ofcure visit were similar between groups, with numerical differences with wide CIs among individual bacterial species (table 2). Per-pathogen eradication rates at the test-of-cure visit in the extended microbiologically evaluable population for common Enterobacteriaceae ranged from 75 0% to 90 9% for ceftazidime-avibactam, and from 60 0% to 88 9% for meropenem; the corresponding eradication rates for P aeruginosa were 42 9% and 40 0%, respectively (table 2). In the extended microbiologically evaluable population, persistence with increasing minimum inhibitory concentrations ( four-times increase) at the end-oftreatment or test-of-cure visit was noted in two (2%) patients in the ceftazidime-avibactam group and 11 (8%) patients in the meropenem group. Multi-locus sequence typing showed that organisms with increasing minimum inhibitory concentrations with the same genotype as the baseline isolate occurred in one patient in the ceftazidimeavibactam group (K pneumoniae), and 11 patients in the meropenem group (nine with P aeruginosa, one with K pneumoniae, one with both P aeruginosa and K pneumoniae). Rates of emergent infections in the extended microbiologically evaluable population were low across both treatment groups (appendix p 73). New infections were identified in five (4%) patients in the ceftazidime-avibactam group and six (5%) patients in the meropenem group. Three (2%) superinfections and three (2%) new infections were identified with P aeruginosa, all in the meropenem group. Overall, one or more adverse events occurred in 302 (75%) patients in the ceftazidime-avibactam group and 299 (74%) patients in the meropenem groups 8 Published online December 15, Vol 376 August 28, 2010

9 (table 3). Adverse events were judged to be treatment related in 66 (16%) patients in the ceftazidime-avibactam group and 54 (13%) patients in the meropenem group. Few adverse events resulted in discontinuation of the study drug (table 3; appendix p 74). Diarrhoea (appendix p 83), hypokalaemia, anaemia, constipation, and vomiting occurred in 5% or more of patients in one or both groups (table 3). No clinically meaningful trends or changes in haematological values, clinical chemistry parameters, coagulation results, or urinalysis results were identified, and no clinical changes of concern were noted for vital signs or electrocardiograms in either treatment group. Serious adverse events occurred in 75 (19%) patients in the ceftazidime-avibactam group and 54 (13%) patients in the meropenem group. The most commonly reported serious adverse events were in the system organ classes of infections and infestations; respiratory, thoracic, and mediastinal disorders; and cardiac disorders. Four patients (1%) in the ceftazidime-avibactam group (and none in the meropenem group) had a serious adverse event that was considered by investigators as possibly related to the study drug: diarrhoea in a 22-year-old man, acute coronary syndrome in a 79-year-old man, subacute hepatic failure in a 33-year-old woman, and abnormal liver function test results in a 22-year-old man. Two of these events led to discontinuation of study drug. All patients had recovered (or the event had resolved) or were recovering at the time of the final protocol follow-up visit. Safety data for the 62 patients with moderate or severe renal impairment excluded after the protocol amendment are in the appendix (p 80). Discussion REPROVE is the first phase 3 study of ceftazidimeavibactam in adults with nosocomial pneumonia (including ventilator-associated pneumonia). To our knowledge it is the first randomised controlled trial to show non-inferiority, compared with a carbapenem, of a new antimicrobial therapy targeting Gram-negative pathogens in this setting. Our results show non-inferiority for the treatment of nosocomial pneumonia caused by ceftazidime-non-susceptible or ceftazidime-susceptible Gram-negative aerobic pathogens. The safety profile of ceftazidime-avibactam in this trial was similar to that of ceftazidime alone and consistent with the profile of ceftazidime-avibactam No new safety concerns were identified, and the overall pattern of adverse and serious adverse events was reflective of the underlying disease and comorbidities in this patient population. Although a numerical difference in the incidence of serious adverse events was noted between the ceftazidime-avibactam and meropenem groups, most were unrelated to study treatment. Per-patient favourable microbiological response rates were generally lower than clinical cure rates, but similar across treatment groups. By contrast with previous Ceftazidimeavibactam (n=405) Meropenem (n=403) All-cause mortality 38 (9%) 30 (7%) Deaths due to disease progression 13 (3%) 8 (2%) Adverse events* Any 302 (75%) 299 (74%) Any with outcome of death 25 (6%) 22 (5%) Any serious adverse events 75 (19%) 54 (13%) Any leading to discontinuation of 16 (4%) 11 (3%) study drug Any of severe intensity 66 (16%) 51 (13%) Adverse events in 2% of patients* Diarrhoea 61 (15%) 62 (15%) Hypokalaemia 43 (11%) 33 (8%) Anaemia 25 (6%) 18 (4%) Constipation 25 (6%) 31 (8%) Vomiting 23 (6%) 22 (5%) Alanine aminotransferase increased 16 (4%) 19 (5%) Aspartate aminotransferase 16 (4%) 17 (4%) increased Oedema peripheral 17 (4%) 15 (4%) Hypertension 14 (3%) 15 (4%) Nausea 13 (3%) 7 (2%) Decubitus ulcer 9 (2%) 6 (1%) Pyrexia 10 (2%) 13 (3%) Hyponatraemia 10 (2%) 6 (1%) Hypotension 10 (2%) 8 (2%) Urinary tract infection 11 (3%) 15 (4%) Abdominal pain 10 (2%) 8 (2%) Pneumonia 10 (2%) 12 (3%) Respiratory failure 10 (2%) 5 (1%) Pleural effusion 9 (2%) 9 (2%) Rash 8 (2%) 13 (3%) Tachycardia 8 (2%) 5 (1%) Cardiac failure 8 (2%) 6 (1%) Atrial fibrillation 5 (1%) 9 (2%) Insomnia 4 (1%) 11 (3%) Data are n (%). Terms defined according to the Medical Dictionary for Regulatory Activity (version 18.1). *Patients with multiple adverse events in the same category were counted only once in that category; patients with adverse events in more than one category were counted once in each of those categories. Excludes patients who died as a result of disease progression. One patient in each group had an adverse event that began before final protocol follow-up and resulted in death after final protocol follow-up; these patients were excluded from this summary. Defined as any event occurring during any study phase that fulfilled one or more of the following criteria: resulted in death, was immediately life-threatening, required in-patient hospitalisation or prolongation of existing hospitalisation, resulted in persistent or significant disability or incapacity or substantial disruption of the ability to conduct normal life functions, was a congenital abnormality or birth defect, was an important medical event that could jeopardise the patient or require medical intervention to prevent one of the outcomes listed above; deaths resulting from disease progression were not counted as serious adverse events. Table 3: Mortality and adverse events up to final follow-up visit (safety population) antibiotic trials (albeit in community-acquired pneumonia), 25,26 no improvement was noted in clinical cure rates for patients who received 24 h or less of Vol 376 August Published 28, online 2010 December 15,

10 previous antibiotics compared with those who did not receive previous antibiotics in either treatment group. These findings might have implications for design of future trials of nosocomial pneumonia, such as potentially extending permitted previous antibiotic use to a period greater than 24 h. An exploratory analysis showed no difference in clinical cure rates between patients who received concomitant aminoglycosides (either 72 h or >72 h of exposure) and those who did not. The favourable clinical outcomes noted with either ceftazidime-avibactam or meropenem without aminoglycoside should be interpreted with caution, however, because REPROVE was not designed to evaluate monotherapy (ie, β-lactam alone) versus combination therapy (ie, a β-lactam plus aminoglycoside). The persistence of an organism with increasing minimum inhibitory concentration with the same genotype as the baseline isolate occurred in 11 patients in the meropenem group (compared with one patient in the ceftazidime-avibactam group), which could affect clinicians choice of empirical antibiotic therapy in the future for patients deemed to be at high risk of recurrence of nosocomial pneumonia, particularly for those with P aeruginosa infections. Subgroup analyses of the primary endpoint showed no treatment differences across various patient subgroups. Despite initial expectations, no difference was observed in clinical cure rates between those with ventilator-associated and non-ventilator-associated penumonia. This finding might be related to improvements in care of patients with ventilator-associated infection, or possibly to the use of a standard comparator (meropenem) in this trial. Furthermore, some patients in whom non-ventilatorassociated pneumonia was diagnosed subsequently required mechanical ventilation. Clinical cure rates were similar in patients with augmented renal clearance, normal renal function or mild impairment, or moderate to severe impairment. During the early stages of REPROVE, results from RECLAIM 1 and 2 became available, 20 suggesting the potential that the per-protocol regimen of ceftazidime-avibactam could be an underdose in patients with moderate or severe renal impairment. Thus, the REPROVE protocol was amended to increase the ceftazidime-avibactam dose in such patients, and patients with moderate or severe renal impairment who received the original dosing regimen were excluded from the main analyses. Efficacy and safety results in these patients were consistent with those in the overall population, but the small size of this subgroup prevents definitive conclusions. The amended dosage modifications are supported by pharmacokinetic pharmacodynamic analyses, 27 and reflect the approved product labelling. 28,29 The mortality associated with nosocomial pneumonia is affected by several factors, and reported frequency varies substantially. All-cause mortality at day 28 (clinically modified intention-to-treat population) were 8% in the ceftazidime-avibactam group and 7% in the meropenem group somewhat lower than some other investigators have reported. 3 However, REPROVE had a representative patient population in terms of ventilator-associated and non-ventilator-associated pneumonia, APACHE II score, and previous antibiotics use within the confines of a clinical study. Patients were not enrolled if they had concurrent morbidities preventing accurate disease assessment, or if they had a high likelihood of dying within the treatment period despite delivery of adequate antibiotics; these exclusions are likely to be reflected in the overall mortality rates. A key limitation of this trial is that we could not establish optimum duration of treatment with either ceftazidimeavibactam or meropenem, and thus it does not provide any additional information that affects the standard of care with respect to these aspects of patient management. Furthermore, various aspects of the design, particularly the duration of study treatment of 7 14 days, although consistent with guidelines available at the start of the study, 30 might not be representative of clinical practice and guidelines, which typically involve antibiotic de-escalation based on culture results. Similarly, the mode of meropenem administration (30 min infusions every 8 h) we used was consistent with the approved label and guidelines, 3,30 but might not reflect how the drug is given now (some institutions give prolonged or continuous infusions). Such design constraints are common in non-inferiority trials, in which careful efforts to avoid confounding the results and falsely concluding non-inferiority are required. Furthermore, the small numbers of patients with bacteraemia limits the applicability of the results to patients with sepsis. In summary, our data support a role for ceftazidimeavibactam as a carbapenem-sparing strategy for nosocomial pneumonia. Contributors AT, NZ, JP, J-FT, MK, and JWC designed and conceived the study. AT, NZ, JP, J-FT, DT, PJL, GGS, and JWC had roles in acquisition, analysis, or interpretation of data. PJL did the statistical analysis. ZC, JS, DT, PJL, and JWC provided administrative, technical, or material support. AT, DT, NZ, JP, J-FT, MK, ZC, and JS supervised the study. ZC, JS, DT, PJL, GGS, and JWC wrote the Article, which was critically revised for important intellectual content by all authors. All authors approved the final version of the Article. Declaration of interests AT received a consultancy fee from AstraZeneca for participating as principal investigator of the study. NZ, JP, J-FT, and MK s institutions received research grant funding from AstraZeneca for the conduct of the study. ZC and JS are employees of AstraZeneca. GGS and JWC were employees of, and shareholders in, AstraZeneca at the time of study completion, and are currently employees of Pfizer. DT was contracted to AstraZeneca from Taylormade Health at the time of study completion, and is a shareholder in AstraZeneca. PJL is contracted to AstraZeneca from the Statistical Services Unit, University of Sheffield. Acknowledgments This study was previously sponsored by AstraZeneca, and is now sponsored by Pfizer. AstraZeneca s rights to ceftazidime-avibactam were acquired by Pfizer in December, Ceftazidime-avibactam is being developed by Pfizer in collaboration with Allergan. Medical writing support, including 10 Published online December 15, Vol 376 August 28, 2010